[Fig 7] The share of non-fossil fuel power generation in Korea
1. Analysis of gross GHG emissions
First, model 1 was analyzed. We computed the augmented Dickey-Fuller statistic to test whether each variable is stable. For these tests, the null hypothesis is that a unit root is present while the alternative hypothesis is that it is not. The test shows that, in the case of levels, the variables E, O, and N have unit roots, but G (GDP) and M do not, at the 95% confidence interval (seeTable3). For the case of differences, as shown in Table4, the null hypothesis
can not be rejected for all variables. Therefore, not all variables have unit roots in differences.
Variables Augmented Dickey-Fuller test
statistic P-value*
GHG emissions (E) -2.024 0.275
GDP (G) -3.361 0.025
Import price of crude oil (O) 0.218 0.967 Heating and cooling degree days (M) -3.457 0.021 Share of non-fossil fuels (N) -2.758 0.082
<Table 3> Test for unit root(level).
Note: Null hypothesis: Each variable has a unit root.
* MacKinnon(1996) one-sided p-values
Variables Augmented Dickey-Fuller test
statistic P-value*
GHG emissions (E) -4.137 0.005
GDP (G) -3.605 0.015
Import price of crude Oil (O) -4.022 0.007 Heating and cooling degree days (M) -3.962 0.009 Share of non-fossil fuels (N) -4.148 0.005
<Table 4> Test for unit root(first difference)
Note: Null hypothesis: Each variable has a unit root.
* MacKinnon(1996) one-sided p-values.
Next, we performed Johansen cointegration tests check for any long-term stable relationship between the variables with unit roots (Johansen, 1988, 1991, 1992; Johansen and Juselius, 1990, 1992, 1994). Table 5 shows the results: the trace statistic, the maximum eigenvalue statistic, and P values. The trace test indicates five cointegrating equations at the 0.05 level, and the max-eigenvalue test identifies four cointegrating equations at the 0.1 level. Therefore, we can conclude that all variables have long-run stable cointegrating relationships.
Unrestricted cointegration rank test (trace) Hypothesized no.
of CE(s) Eigenvalue Trace statistic 0.05 Critical value Prob.**
None * 0.833 103.689 69.819 0.000
At most 1 * 0.725 67.918 47.856 0.000 At most 2 * 0.669 42.102 29.797 0.001 At most 3 * 0.493 20.004 15.495 0.010 At most 4 * 0.275 6.432 3.841 0.011 Note: Trace test indicates five cointegrating equations at the 0.05 level.
* denotes rejection of the hypothesis at the 0.05 level.
** MacKinnon-Haug-Michelis(1999) p-values.
Unrestricted cointegration rank test (maximum eigenvalue) Hypothesized no.
of CE(s) Eigenvalue Max-eigen
statistic 0.05 Critical value Prob.**
None * 0.833 35.771 33.877 0.029
At most 1 * 0.725 25.816 27.584 0.083 At most 2 * 0.669 22.098 21.132 0.037 At most 3 * 0.493 13.572 14.265 0.064 At most 4 * 0.275 6.432 3.841 0.011 Note: Max-eigenvalue test indicates one cointegrating equation at the 0.05 level.
* denotes rejection of the hypothesis at the 0.05 level.
** MacKinnon-Haug-Michelis(1999) p-values.
<Table 5> Test for cointegration(level).
As shown above, these models can be analyzed by FMOLS on their long-term aspects since these variables are cointegrated, although some variables have unit roots. This paper considered four models to analyze gross GHG emissions.
The FMOLS regression results on long-term equilibrium relationships (see Table 6) show how each variable can affect gross GHG emissions. Model 1 used (GDP), (import price of crude oil), (heating and cooling degree days), and (share of non-fossil fuels) as the exogenous variables. The exogenous variables used were in
Model 2, in Model 3, and in Model 4. We can verify, using different combinations of variables, the effects of each variable. However, this study focuses on Model 1, because the coefficients of all variables in the model are statistically significant.
Specifically, the variables are significant at the 95% confidence interval. The result of Breusch-Godfrey serial correlation LM test shows no serial correlation.
GDP has the largest effect on gross GHG emissions. This is followed by heating and cooling days. According to the results of this research, a 1% increase in GDP and in heating and cooling degree days raises gross GHG emissions by 0.598% and 0.463%, respectively. In contrast, the share of non-fossil fuels and the import price of crude oil have a reducing effect on gross GHG emissions in Korea. A 1% increase in the share of non-fossil fuels and in the import price of crude oil reduces gross GHG emissions by 0.162% and 0.017%, respectively.
However, the GHG reduction effects of the import price of crude oil is minimal, considering that the coefficients of the import price are extremely low, although statistically significant.
Variables
Model 1 (E,G,O,M,N)
Model 2 (E,G,O,M)
Model 3 (E,G,O,N)
Model 4 (E,G,M,N) Coefficie
nt
Std.
rror
Coefficie nt
Std.
Error
Coefficie nt
Std.
Error
Coefficie nt
Std.
Error GDP (G) 0.598*** 0.021 0.658*** 0.028 0.552*** 0.041 0.561*** 0.015 Import price of
Crude Oil (O) -0.017* 0.009 -0.034** 0.015 0.003 0.018 Heating and
cooling degree days (M)
0.463*** 0.065 0.597*** 0.099 0.410*** 0.069 Share of
non-fossil fuels (N)
-0.162*** 0.050 -0.313*** 0.095 -0.202**
* 0.054
R-squared 0.990 0.987 0.980 0.989
Durbin-Watson d
statistic 1.799 1.832 1.073 1.355
<Table 6> Estimated models of gross GHG emissions(FMOLS).
The above analysis considers both heating degree days and cooling degree days. However, the effects of the heating degree days and cooling degree days are expectedly different. Therefore, two additional analyses were added to this research. <Table 7> shows the results on heating degree days and Tables 8 on cooling degree days. From the FMOLS analysis, the coefficient of heating degree days is 0.333, as shown in <Table 7>, and that of cooling degree days is 0.155, as shown in <Table 8>. These results show that heating degree days affect GHG emissions in Korea more than cooling degree days do. This result is consistent with the common understanding that energy consumption is greater on heating degree days than on cooling degree days.
(E,G,O,MH,N) (E,G,O,MH) (E,G,MH, N) Coefficient Std.
Error Coefficient Std.
Error Coefficient Std.
Error
GDP (G) 0.588*** 0.028 0.666*** 0.037 0.560*** 0.019
Import price of crude oil (O) -0.012 0.012 -0.033 0.019
Heating degree days (MH) 0.333*** 0.080 0.484*** 0.120 0.298*** 0.079 Share of non-fossil fuels (N) -0.203*** 0.066 -0.231*** 0.068
R-squared 0.988 0.983 0.980
Durbin-Watson d statistic 1.242 1.580 1.002
<Table 7> Estimated models of gross GHG emissions(FMOLS).
(E,G,O,MC,N) (E,G,O,MC) (E,G,MC, N) Coefficient Std.
Error Coefficient Std.
Error Coefficient Std.
Error
GDP (G) 0.562*** 0.032 0.649*** 0.042 0.551*** 0.021
Import price of crude oil (O) -0.003 0.014 -0.022 0.021
Cooling degree days (MC) 0.155*** 0.050 0.201** 0.072 0.154*** 0.050 Share of non-fossil fuels (N) -0.256*** 0.078 -0.264*** 0.077
R-squared 0.983 0.976 0.983
Durbin-Watson d statistic 2.004 1.639 1.954
<Table 8> Estimated models of gross GHG emissions(FMOLS).
<Table 9> shows the results of VECM models in equation (3). shows the error correction terms and statistically significant at a 1% level. In the short run, GDP, Heating and Cooling degree days, and the share of non-fossil fuels clearly affected on GHG emissions. The coefficients of Heating and Cooling degree days and share of non-fossil fuels are 0.62 and -0.38 and statistically significant at a 1% level. GDP also increases on the GHG emissions in the short run and statistically significant at a 1% level. But the oil price did not affected on the GHG emissions in the short run because the coefficient is negative but is not statistically significant.
Coefficient Std.
Error Coefficient Std.
Error Coefficient Std.
Error
-0.80 0.86 -0.88 1.52 -7.42** 4.09
0.58*** 0.17 0.73*** 0.30 -1.10 0.79
-2.31*** 0.57 -2.19** 1.01 -4.84* 2.70
-1.57*** 0.48 -1.37 0.84 -3.18 2.26
1.61*** 0.41 1.55** 0.72 3.86* 1.94
1.19*** 0.38 1.01 0.67 2.02 1.81
0.03 0.06 0.04 0.10 0.00 0.28
0.06 0.05 0.11 0.08 -0.07 0.23
0.62*** 0.17 0.47 0.30 2.28* 0.80
-0.38*** 0.13 -0.21 0.23 -0.16 0.60
R-squared 0.84 R-squared 0.58 R-squared 0.80